The connect-ad study: communication and network navigation for effective care teams in alzheimer's disease and alzheimer's disease related dementias (ad/adrd)

The National Institute on Aging awarded the University of California, Los Angeles $534,287 on September 30, 2025, under the Aging Research program (CFDA 93.866) to develop the CONNECT-AD Study, a multisite investigation of electronic health record communication networks in Alzheimer's disease and related dementia care teams.

The study uses social network analysis, machine learning-assisted visual analytics, and qualitative methods to analyze EHR communication structure data across three University of California Health Systems that all operate Epic systems. The research builds on prior work in cancer care coordination to address fragmented communication among multiple providers treating AD/ADRD patients. Pilot analyses identified 30,523 AD/ADRD patients with over 2.7 million patient and caregiver portal messages and 108,709 healthcare providers asynchronously accessing their digital records. The work focuses on information sharing via EHRs as a modifiable dimension of team communication and incorporates patient and caregiver communications via EHR patient portals alongside provider-to-provider coordination. Primary aims include developing the first multisite database of EHR multiteam system communication in AD/ADRD care with novel social network measures to quantify communication patterns.

Place of performance is Los Angeles, California. The period of performance runs from September 30, 2025, through August 31, 2027. The University of Iowa serves as a sub-recipient, leading data curation and statistical analyses under Dr. Daniel Sewell as site principal investigator. The University of Iowa's responsibilities include curating datasets into structured relational databases, developing measures investigating multiteam system communication through the EHR, examining relationships between EHR communication and quality outcomes in AD/ADRD care, manuscript preparation, and participation in regular research team meetings.

Abstract

Alzheimer's disease and alzheimer's disease related dementias (ad/adrd) are leading causes of morbidity, mortality, and excess healthcare costs. Care for ad/adrd patients requires multiple providers, often hindered by fragmented communication. As treatment options continue to expand and an aging population with frequent comorbidities, challenges in care coordination will multiply and represent an urgent need in healthcare. In our prior work we have found that digital tracings of communication network structures identified within the electronic health record (EHR) predict cancer patients' outcomes (e.g., ED visits, mortality), hence we are developing scalable EHR tools to improve cancer patients' care. Our results in cancer provide a strong foundation for studying ad/adrd patients, given similarities in the complexity of care coordination across multiple providers and specialties. This study of persons with ad/adrd expands on our prior work in cancer by factoring in the longer-term nature of ad/adrd patient care and adding a significant new dimension to the care team: patient and caregiver communications via EHR patient portals. We will leverage social network analysis, machine learning (ML)-assisted visual analytics, and qualitative methods to study EHR communication structure data at three university of california health systems that all use epic. Pilot analyses identified n=30,523 ad/adrd patients with over 2.7 million patient/caregiver portal messages and 108,709 healthcare providers asynchronously accessing their digital records. We focus on one modifiable dimension of team communication, information sharing via ehrs. Aim 1: develop the first multisite database of EHR multiteam system communication in ad/adrd care, incorporating novel social network measures to describe and quantify within- and between-group collaborative patterns in intricate EHR interactions. Aim 2: analyze the association between targeted EHR communication structures and quality outcomes through social network analyses, focusing on preventable ED visits and unplanned hospitalizations in ad/adrd care. Aim 3: utilize machine learning-assisted techniques to characterize and assess collaborative care dynamics, identify EHR communication structures associated with poor quality outcomes, and predict patient outcomes and events. Aim 4: apply qualitative methods to deepen our understanding of the findings from aims 1-3, exploring how and why communication structures and teaming patterns impact collaborative care delivery. Multisystem healthcare teams and communication structures need to be understood and then designed much more intentionally and based on evidence. Our innovative methods will provide data to help US improve our fragmented healthcare system and ensure quality care.